Language impairments and communication disorders are considered cardinal features of autism. The fact that language impairment is a hallmark of autism does not mean, however, that there is one "language syndrome" exhibited by all autistic individuals. Individual variation is substantial. Higher functioning individuals exhibit difficulties in pragmatic functions and complex verbal processing. Lower functioning individuals (approximately 30 percent) are mute. Very little research has been aimed at the identification of early speech procursors that might correlate with later high- vs. Low- functioning linguistic behavior in autism. Such research is crucial to our understanding of the nature of the discorder and for developing effective strategies for early intervention. Over the past 20 years, the research focus in the P.I.'s laboratory has been early language development in typically-developing infants. This research has identified five important components of early linguistic competence that hold promise as predictor measures of language impairment in autism. A unique strength of this proposal is that these state-of-the-art speech measures will create a "speech profile" for each individual child. The five measures comprising the profiel where chosen to provide breath and depth encompassing five critical processes in speech-language acquisition: (1) Basic phonetic discrimination - do young children with autism lack the basic ability to differentiate speech sounds, suggesting primary language processing deficits? (2) Prototype formation - Do the language deficits in children with autism stem from a failure to mentally "map" socially-provided linguistic imput? (3) Auditory preference for speech - When given a choice, do children with autism choose to listen to human speech vs other control sounds? (4) Vocal imitaiton - How is echolalia different from the social-motivated vocal imitation observed in typically-developing children? (5) Cross-modal (auditory-visual) speech perception - Do children with autism cross-modally match auditory and visual speech? Data on these five speech measures collected at 3-4 years of age (Time 1) on the cohort of children with autism defined for the project will be compared to identical measures taken on developmentally-matched populations of typically-developing and mentally-retarded children. In addition, the early speech measures will be correlated with; (a) other behavioral and neuropsychological measures taken at Time 1 and Time 2 (Project I and III) and (b) measures of brain structure and metabolism (magnetic resonance imaging and spectroscopy) at Time I and Time 2 (Project IV).